. 2023 May 8;52(5):1605–1632. doi: 10.1007/s10936-023-09945-4

Table 4.

Principal component analysis was conducted to explore narrative data structure at T0 (pre-training) and T1 (post-training)

	T0—Pre-training				T1—Post-training
	Pattern matrix^a		Structure matrix		Pattern matrix^b		Structure matrix
	Component		Component		Component		Component
	1	2	1	2	1	2	1	2
MLU	.858	.194	.868	.235	− .632	.074	− .625	.013
Omissions of morphosyntactic information	− .907	− .071	− .911	− .114	.893	− .044	.889	.041
Complete sentences	.875	.054	.878	.096	− .828	.020	− .826	− .059
Cohesive errors	− .666	.432	− .645	.400	.814	.200	.833	.278
Coherence errors	.323	.890	.365	.905	− .040	.831	.040	.827
LIUs	.108	− .962	.062	− .957	− .041	− .895	− .127	− .898

Extraction method: Principal component analysis. Rotation method: Oblimin with Kaiser normalization

^aConvergence for rotation performed in 5 iterations

^bConvergence for rotation performed in 3 iterations

MLU Mean length of utterance; LIUs Lexical Information Units